Towards a personalized digital patient for diagnosis and therapy guided by image

Recent advances in computer science and medical imaging allow the design of new computational models of the patient which are used to assist physicians. These models, whose parameters are optimized to fit in vivo acquired images, from cells to an entire body, are designed to better quantify the observations (computer aided diagnosis), to simulate the evolution of a pathology (computer aided prognosis), to plan and simulate an intervention to optimize its effects (computer aided therapy), therefore addressing some of the major challenges of medicine of 21(st) century.     

Medical data base system with an ability of automated diagnosis.

We carried out an experiment on a medical information system in which a clinical data base is combined organically with computer programs for automated diagnosis. In this system, the parameters for automated diagnosis are devised to be renewed as the contents of the data base (patient's information) increase. This system can be regarded as a data base possessing a kind of diagnosing ability which grows up with time. We have named this system "Intelligent Data Base". The algorithm for computer diagnosis used in this study is based on maximum likelihood method, and each likelihood is weighted with a prior probability of each disease. The discrimination efficiency of this method is logically equal to that of the Bayes rule. First 27 cases were learnt by the system and correct diagnosis was obtained in 78% of the cases. When cases for learning increased to 82, the percentage of correct diagnosis was improved to 95%.     

Detection of renal allograft rejection by computer.

A computer program incorporating an adaptation of a statistical method, the multiprocess Kalman filter, was used to detect changes in trends of plasma creatinine and urea concentrations. In 28 recipients of renal allografts a definite deterioration in renal function was identified retrospectively on 32 occasions by an experienced renal physician independently of the statistical analysis. The computer identified 31 of these 32 episodes using creatinine and urea results, and 29 using creatinine alone. Dysfunction was identified by the computer significantly earlier (p less than 0.05) than by the clinician and a median of one day earlier (p less than 0.02) than treatment was actually initiated. The computer identified dysfunction on 11 out of 1259 days when the clinician did not suspect rejection. These 11 episodes may have had a pathological importance, though no clinical diagnosis was made. This computer method is useful for immediate analysis of incoming results and for timing events either prospectively or retrospectively.     

The suitability of digital colposcopy for telematic applications]

Since carcinoma of the cervix is one of the most common cancers in women, screening of the cervix has acquired considerable importance. Colposcopy is a simple diagnostic method of detecting suspicious changes at an early stage. Shortcomings of this method are its low specificity and high inter- and intra-observer variability. A clinical pilot study was therefore carried out to investigate the advantages of a digital colposcopic system comprising a binocular colposcope coupled to a CCD camera and a computer. The aim of the study was to evaluate the reliability of diagnostic findings of the cervix obtained with digital colposcopy in comparison with standard binocular colposcopy, and to assess its suitability for telematic applications (teleconsultation, telediagnostics, treaching). A total of 315 patients were examined and statistically analysed. The patients were first submitted to a conventional colposcopic examination and a diagnosis was established. During the colposcopic examination camera images were stored on a computer, on the basis of which a second physician experienced in colposcopy reviewed the initial diagnosis. The primary and secondary findings of each patient were classified into 4 categories and compared following the Rome classification system. Agreement between the primary and secondary diagnosis was established in 69% of the cases (kappa = 0.60 +/- 0.03). No bias was observed in terms of under- or overrating. The percentage of non-assessable colposcopic examinations was 9.2%. Digital colposcopy is therefore suitable for reproducing diagnostic findings on the computer, given adequate digital image quality and a suitable classification model. The method has clear advantages with regard to follow-up, internal quality control of the diagnosis, and the training and further education of physicians and students. In the future, telecolposcopy may open up new opportunities in gynaecology.     

Doctors as Decision-makers: A Computer-assisted Study of Diagnosis as a Cognitive Skill

When viewed as a sequence of decisions clinical diagnosis becomes amenable to detailed investigation in terms of standard statistical concepts. A study of six clinicians diagnosing identical sets of cases of non-toxic goitre is used to illustrate an objective technique for studying the diagnostic process with the aid of a digital computer. Considerable variation in clinicians'' routes to correct diagnosis is shown when these routes are compared in detail by five statistical measures related to the effective use of the information available to the clinicians. For rapid analysis of diagnostic skill two visual methods are presented. These can be developed for teaching undergraduates the interpretative skills involved in diagnosis and for studying such skills in experienced clinicians.     

Diagnosis of gastric cancer.

A prospective comparison was made of the accuracy of different diagnostic methods for gastric cancer. The basis of the study was a consecutive series of 113 patients thought to have gastric pathology; cancer was the final diagnosis in 32. Endoscopy and radiology were the most accurate investigations, whereas biopsy, cytology, and clinical examination gave disappointing results. A wide range of clinical features and laboratory investigations were studied in all patients in an attempt to identify criteria suggestive of malignancy. Multifactorial computer analysis of these investigations failed to improve upon the radiological diagnosis. A systemic approach designed to make optimal use of limited endoscopic and histopathological resources in the diagnosis of gastric lesions is presented.     

Some mathematical aspects of the diagnostic process. II: A mathematical model of electrocardiographic diagnosis

In a previous paper (Bartholomay, 1971), a general mathematical model of the medical diagnostic process was described. The present paper amounts to a relization of that process in terms of conventional 12-lead electrocardiographic diagnosis as enunciated by Dr. Harold D. Levine (1966) in the course of a collaborative study by Dr. Levine and the present author at the Peter Bent Brigham Hospital of the Harvard Medical School between 1963 and 1966. The main details of the cognitive component of that model are described in detail here. The model has been programmed onto a computer system consisting of an analog-digital converter and general purpose digital computer and amounts to a simulation of Dr. Levine’s electrocardiographic analysis procedure.     

Comparison of the diagnostic methods used in maxillofacial trauma

A prospective study was performed to evaluate the diagnostic methods used in acute maxillofacial trauma. Clinical examination, routine facial x-rays with linear tomography, and computer tomography were compared in 49 patients. Computer tomography was found to be the most accurate test in the diagnosis of facial bone injury, especially complex fractures. Computer tomography also provided valuable information regarding soft-tissue injury of the face. The radiation doses of linear tomography and computer tomography were calculated to be below the level known to cause cataract formation. Computer tomography is a safe, reliable adjunct in the diagnosis of acute maxillofacial trauma and should be strongly considered after initial screening measures are completed.     

Computer aided diagnosis of acute abdominal pain: a multicentre study.

A multicentre study of computer aided diagnosis for patients with acute abdominal pain was performed in eight centres with over 250 participating doctors and 16,737 patients. Performance in diagnosis and decision making was compared over two periods: a test period (when a small computer system was provided to aid diagnosis) and a baseline period (before the system was installed). The two periods were well matched for type of case and rate of accrual. The system proved reliable and was used in 75.1% of possible cases. User reaction was broadly favourable. During the test period improvements were noted in diagnosis, decision making, and patient outcome. Initial diagnostic accuracy rose from 45.6% to 65.3%. The negative laparotomy rate fell by almost half, as did the perforation rate among patients with appendicitis (from 23.7% to 11.5%). The bad management error rate fell from 0.9% to 0.2%, and the observed mortality fell by 22.0%. The savings made were estimated as amounting to 278 laparotomies and 8,516 bed nights during the trial period--equivalent throughout the National Health Service to annual savings in resources worth over 20m pounds and direct cost savings of over 5m pounds. Computer aided diagnosis is a useful system for improving diagnosis and encouraging better clinical practice.     

A video digitizer for analysis of trunk deformity in scoliosis

Scoliosis is a deformity characterized by lateral curvature of the spine and accompanied by axial rotation of the vertebrae; it often causes varying degrees of trunk deformity. Research has indicated that topographic techniques can be used to describe the disorder and monitor its progression. A video image acquisition system has been designed which reduces the time required to quantify topographic details of the trunk and aids in the diagnosis, monitoring and research of scoliosis. This system integrates the capability of large, expensive grey-scale image acquisition equipment into a small, low-cost diagnostic imaging tool using current technologies and design techniques. The video digitizer accepts a standard NTSC monochromatic video signal as input and the unit is connected to a computer via an EEE-488 bus from which the ditigizer is controlled. The digitizer samples the video signal in real time using a high-speed flash converter controlled by an application-specific integrated circuit; the digital samples are stored in memory until the host computer requests that the information be transferred.     

基于专家数据库的人体信息诊疗仪的研制

本文介绍了我们研制的一种基于电脑专家数据库的人体信息诊疗仪。融合了先进的激光技术、电子技术、计算机技术和人体信息诊疗学于一体。该仪器由两部分组成：（1）可自动诊断的电脑专家数据库;（2）半导体激光针灸部分。     

Real-time fuzzy-knowledge-based control of Baker's yeast production

A real-time fuzzy-knowledge-based system for fault diagnosis and control of bioprocesses was constructed using the object-oriented programming environment Small-talk/V Mac. The basic system was implemented in a Macintosh Quadra 900 computer and built to function connected on line to the process computer. Fuzzy logic was employed in handling uncertainties both in the knowledge and in measurements. The fuzzy sets defined for the process variables could be changed on-line according to process dynamics. Process knowledge was implemented in a graphical two-level hierachical knowledge base. In on-line process control the system first recognizes the current process phase on the basis of top-level rules in the knowledge-base. Then, according to the results of process diagnosis based on measurement data, the appropriate control strategy is subsequently inferred making use of the lower level rules describing the process during the phase in question. (c) 1995 John Wiley & Sons, Inc.     

Use of computer program for diagnosing jaundice in district hospitals and specialized liver unit.

A computer-assisted model for diagnosing jaundice has been adapted for use on the University of London C.D.C. 7600 computer via an on-line terminal at King''s College Hospital to provide a rapid turn-round time. The model was used prospectively in the diagnosis of 219 patients--135 seen in a specialized liver unit and 84 seen in one of four district hospitals in south-east London--with an overall accuracy in distinguishing among 11 different causes of jaundice of 69% and 62% respectively. These figures rose to 77% and 88% respectively when only those patients in whom the final diagnosis reached a "certain" probability were considered. When used to distinguish between a medical and a surgical cause of jaundice the accuracy was 86% in the liver unit and 77% in the district hospitals, rising to 95% in both series for those with a diagnosis of certain probability. The proposed improvements to the model--namely, the use of two deparate data bases and more diagnoses within the matrix--should be improve the accuracy even further. In practice the rapid feedback to the clinicians looking after patients provided help in managing difficult cases.     

Simplified computer-aided diagnosis of acute abdominal pain.

A simplified version of a system for computer-aided diagnosis of acute abdominal pain has been tested by "new" personnel unfamiliar with the previous system. After a two-month learning period the system proved more accurate in its diagnoses than the unaided clinician, and during the first five months of using the system the unaided clinicians'' accuracy rose from 73% to 84%. When computer "feedback" was withdrawn the clinicians'' diagnostic accuracy reverted towards the previous, "unaided" level. These findings further validate the concept of the computer as a potentially valuable diagnostic aid but indicate that a training period and computer feedback are important factors in its 4 use.     

Human versus machine: A comparison of a computer ‘expert system’ with human experts in the diagnosis of vaginal discharge

A model system has been designed which generates ‘case’ of vaginal discharge. Each such case is presented to a human for diagnosis, and this is then compared with a computer diagnosis using two forms of Bayes' theorem. Six subjects (2 medical; 4 non-medical) participated in the trial and each examined 100 successive ‘cases’. When the human had forewarning of the trial and full access to the knowledge-base their performance was superior to that of Bayes' theorem using positive features only and equivalent to that using both positive and negative features. When the trial was repeated without forewarning the human performance was markedly inferior to that of the machine. It is concluded: (1) that human and computer-aided diagnosis can be of approximately equal efficiency for complex and non-definitive data; (2) that the imperfections of human memory give an obvious potential advantage to the machine in this type of situation.     

Planning for the information age: a survey of microcomputer use in a faculty of health sciences.

Microcomputers can greatly enhance information processing by clinicians and improve the quality of health care. We surveyed 983 full- and part-time faculty members to assess the state of microcomputer use in the Faculty of Health Sciences at McMaster University, Hamilton, Ont. The ratio of faculty members to microcomputers was close to 1; however, 29% of the full-time and 52% of the part-time members who responded indicated that they did not use a computer. Among those who did, the range of applications was generally limited. There was no mention of more advanced uses such as diagnosis, treatment and patient records. Only about 30% of the respondents had taken a computer course, but all indicated a desire to take courses (on average, three of the seven listed in the questionnaire). Our results showed an extensive but unequal distribution of microcomputers and revealed the need for planning and education to put them to optimal use.     

Adrenal imaging

The differentiation of adrenal nodules is wide and varies from primary benign neoplasia, through hormone secreting lesions, to primary and secondary malignant masses. With the rapid development of cross-sectional imaging, incidental detection of adrenal nodules has become an everyday practice, leaving clinicians with the necessity of further investigation. In this article, we present the current possibilities of adrenal gland imaging and we propose a diagnostic schema for differential diagnosis of incidentaloma. Non-contrast enhanced computer tomography (CT) is a modality of choice used for the differential diagnosis of adrenal lesions. It allows the detection of 95% of adrenal masses, and the characterisation of most of them. Magnetic resonance imaging (MRI) is a good modality for cases in which CT examination cannot determine the character of an adrenal tumour. Nuclear medicine study with the use of Iodine-131 meta iodobenzylguanidine(MIBG) is helpful in the diagnosis of pheochromocytoma. Positron Emission Tomography-Computed Tomography (PET-CT) is considered a useful method in patients with a known malignancy history. Ultrasound has a low sensitivity for the detection of small lesions and is not capable of reliable characterisation of visualised masses. However, this technique plays an important role in the follow-up of non-hypersecreting adrenal lesions.     

Repeatability of nerve conduction measurements derived entirely by computer methods

Background  

Nerve conduction studies are an objective, quantitative, and reproducible measure of peripheral nerve function and are widely used in the diagnosis of neuropathies. The purpose of this study is to determine the reliability of nerve conduction parameters derived entirely from computer based data acquisition and waveform cursor assignments and to quantify the relative contributions of test variability sources.     

Polarity coincidence correlation coefficient and signal energy ratio of the ongoing EEG activity. II. Brain tumors.

The asymmetry of the EEG in 35 patients with brain tumor was analyzed using a special purpose computer which provided two measures: Polarity Coincidence Correlation Coefficient (PCC) and Signal Energy Ration (SER). Significant differences were found between this group and a control group of normal subjects. A discriminant function was calculated using both measures and gave an overall separation accuracy of 87% in the sample studied. It was concluded that the method described offers great utility in routine screening and diagnosis of brain tumors.